Bessent open to discussing shared AI risks with China ahead of summit

US Treasury Secretary Scott Bessent said Wednesday (local time) that the United States is open to discussing shared AI risks with China, as he prepares for high-level negotiations ahead of a bilateral summit. The remarks come as President Donald Trump has repeatedly dismissed calls to slow AI development, citing fierce competition with China — raising questions about whether the two sides can find common ground on AI safety.

Speaking to Axios, Bessent said the United States remains "the leader in AI" and that Washington is "open to discussions to avoid risks that both sides face in common and to prevent the AI systems of the two countries from becoming siloed from each other." He added that talks between the two sides would likely cover both open-weight and closed-weight AI models.

Open-weight models make a model's parameters publicly available and allow users to modify them for custom applications. Chinese AI companies, including DeepSeek, have primarily developed open-weight models.

Closed-weight models, by contrast, keep those parameters proprietary and allow access only through services provided by the developer. Leading US AI companies such as Anthropic and OpenAI maintain closed-weight models.

AI company CEOs have long called for US-China cooperation on AI safety standards.

Bessent is set to meet Chinese Vice Premier He Lifeng in New York on Sunday. The talks, expected to last several hours, will cover AI as well as trade, rare earths and other economic issues, and are seen as a final round of agenda-setting ahead of a US-China summit at the White House on Sept. 24.

Axios said the meeting between the two economic chiefs "could lay the groundwork for discussions on US-China AI safety cooperation."